Delayed maintenance policy optimisation based on (X)over-bar control chart

Data sharing between statistic process control (SPC) and condition-based maintenance is valuable and the joint optimisation has been studied, mostly focusing on the SPC control chart limits. Traditionally, maintenance is taken as a response to the control chart alarms, as soon as the alarm is released. This may not be a good decision due to the existence of false alarms and the loss of production interruptions. So this paper proposed a delayed maintenance policy. This policy allows a delay time for the detection and maintenance after an alarm. The operational state probabilities during the delayed period are estimated by Bayesian theory, and a Markov model is built for the monitoring-maintenance process. The model is validated by a Tecnomatix-based simulation, and then used to optimise the average delay time as well as the sampling parameters. Numerical results show that the improvements do exist in some cases, but it depends on the production conditions. Suggestions about when to perform delayed maintenance are also given through factorial analysis.